TG2 will be a java rewrite of TG1. It will make use of standard components where possible
See also: TermGenie
Meeting in Berkeley June 19 2011 to discuss UI framework
We will use the OWL API (+OWLTools wrapper)
In progress (Heiko to add link to source)
- language for generating class expression from template
- constraint rules for excluding non-sensical expressions. Options:
- encode as OWL (higher reasoner penalty)
- hardcoded java rules (not scalable, no declarative)
- declarative rule language
New compositional terms have auto-generated
- names and synonyms
- textual definitions
We would have some declarative language for specifying these, but some could be hardcoded for now.
TG1 used custom prolog reasoning to efficiently place new terms in the hierarchy. TG2 will use standard OWL reasoners.
We are still evaluating options
- Pellet has a useful incremental reasoning feature, but may be awkward to use. Pellet can be configured to run faster (See RobH's paper on EL-Vira)
- JCel should be fast
- CB should be very fast, but harder to install/use
TG1 wrote out 3 diff files. A separate script is executed by the gatekeeper.
We can retain this model but ultimately TG will have direct cvs access and the gatekeeper should be able to release all changes to the live ontology
TODO: ask owlapi list for standard way of representing transactions
- basic authentication
- some metadata generated from usernames